We’re still at the start of a new decade, but enough time has passed to make and break brittle resolutions. But when it comes to your career, what [data] resolutions have you made and kept? We asked the speakers of the upcoming DSS ATX about what matters to them in 2020: keeping on top of data trends, hiring and continuing education, and being an excellent manager.
Predicting the Next Big Thing for 2020
There’s a lot to be said about the current trajectory of AI and ML, and predictions about the future of these disciplines are all over the map. Many old issues will continue to be relevant in 2020, starting with data enrichment. “As organizations start to recognize the power of external data to move the needle and have an impact on business, they will begin looking for better data instead of better model,” said Maor Shlomo, CEO of Explorium. “In 2020, data enrichments from external data sources will take center stage as organizations open up to data sources they never thought about because they were focused only on internal data. These enriched actionable insights will enable better decisions based on better data and open organizations up to new lines of business and increased revenue.”
Best practices will also take center stage in 2020. “Verifiability and explainability are topics that are super hot,” said Raffael Marty, VP Research and Intelligence at Forcepoint X-Labs. “Google just announced some explainability features for their ML, but it’s something that is still super early. We need a way to understand what algorithms actually find to validate things.” And there’s still the problem of exploring what’s under the hood as well. “We need to better understand bias in ML,” said Marty. “Bias in data, in the algorithms, in the algorithmic parameters, and even in the interpretations and presentation of the data to users is rendering a lot of ML useless.”
Of course, we’ll also have some new shiny tools in 2020 as well. “In 2019, we saw a lot of new ML libraries emerging and becoming more mature,” said Priscilla Boyd, Senior Manager, Data Analytics at Siemens Mobility. “I suspect next year will be all about feature engineering and making parts of the process as automated as possible. We’ve seen some applications attempting it already and I’m sure there will be more coming soon.” Michael Zelenetz, Analytics Project Leader at New York-Presbyterian Hospital, has a different take: “I think this will be the year of the knowledge graph. Knowledge graphs have been bubbling under the surface of the mainstream for years and I think this year we’ll see them integrate more fully and seamlessly to improve other ML and NLP applications.”
Despite the divergence of predictions, the writing’s on the wall: AI and ML are expanding and changing fast in 2020 and beyond.
How to hire in 2020
One thing is for certain in 2020 — just about everyone is hiring. There’s been a lot of ink spilled about finding the perfect data scientist to add to a team. What are the skills that really matter in 2020?
Some of them are the same as they’ve always been — companies want data scientists with domain knowledge. “As data science becomes more mainstream, we’re looking for individuals with in-depth applied mathematics skills and with AI experience coupled with the domain knowledge that we need in transportation,” said Boyd. Marty takes a similar approach: “We are looking a lot at NLU — natural language understanding to assess and understand human behavior. Therefore, anything natural language related and anything belief network oriented are skills that are in high demand for us.”
But like any other career, sometimes it’s a person’s disposition that makes them perfect for the position. “We focus less on hiring for individual technological skills, as those demands change, and focus more on finding people who can learn and adapt. We look for people who love to learn and can pick up new skills as needed,” said Zelenetz. “Another important skill we look for is the ability to communicate with other teams both technical and non-technical. Being able to understand the problems our stakeholders face and then to explain our findings or solution is critical.”
How to grow in 2020
With so many companies looking for so many qualities in data scientists, it’s essential to keep skills sharp. Here’s where our speakers agree: it’s important to keep a sense of curiosity and read, read, read. “I read a lot of papers,” said Zelenetz. “Blogposts, tweets, and I listen to podcasts to keep up to date with developments in the field.” “There are so many online resources out there,” said Boyd. “For me, the key is to reserving and dedicating time to sharing (and obtaining) the knowledge with our teams in training courses or tutorials as they are taken.”
Often that means taking a hand-on approach. “To keep my skills sharp I often write small demos,” said Zelenetz. Jacob Claussen, Manager, Analytics/Data Science at Zynga, agreed: “I still code heavily in my free time (what little bit of it I have these days), but I also try to be an active part of the early research and prototyping phase with team projects. I’ve continued to expand my knowledge of methods and strategies used by others in the industry in order to maintain some thought leadership.”
How to lead in 2020
Most data scientists work in a collaborative environment. Having an effective manager can make or break your chance of success on a team. “I’ve been blessed to have had great managers,” said Zelenetz. “What made them so effective was that they were able to help filter requests and to protect our time. They gave us a long leash to explore problems but kept us focused on our groups’ strategic goals to keep us pointed in the right direction.”
A manager’s philosophy goes a long way here. “Being analytical, understanding the problem domain and problem statement in depth is critical to success,” said Boyd. “Additionally, it’s important to understand and sympathize with the data science processes that we have to undertake. We work with agile and design thinking-focused processes, which means we may sometimes fail. And that’s OK. As a manager, it’s important to know that failure is acceptable and that everything we learn, we should apply in future to avoid it from happening again.” Marty agreed, and offered these questions as checkpoints: “What could go wrong? How can you prevent it early? How do you measure success?“ Setting Key Performance Indicators can help you benchmark success.
Ultimately, a manager is just as accountable to her team as she is to her superiors. “The most effective managers are able to properly assess their team members’ current needs and match them with an effective management style. I have seen teams fail miserably when talent is over/underestimated or managers can’t adapt their style to the situation. This is essential to optimizing team resources on projects and supporting overall team growth and happiness,” said Claussen. “Another important part of team growth and happiness is opportunity. It’s imperative to define new challenges for people and give them the tools to succeed. This includes teaching them and/or offering resources as necessary, giving them chances to fail and learn, and supporting/championing their work throughout the process from inception to conclusion.”
Find out more
Curious for more?
Don’t miss the next → Data Science Salon in Austin, February 18–19, 2020.